Reformulation of the pressure-dependent recruitment model (PRM) of respiratory mechanics

Virtual patient framework for the testing of mechanical ventilation airway pressure and flow settings protocol

BACKGROUND AND OBJECTIVE

Model-based and personalised decision support systems are emerging to guide mechanical ventilation (MV) treatment for respiratory failure patients. However, model-based treatments require resource-intensive clinical trials prior to implementation. This research presents a framework for generating virtual patients for testing model-based decision support, and direct use in MV treatment.

METHODS

The virtual MV patient framework consists of 3 stages: 1) Virtual patient generation, 2) Patient-level validation, and 3) Virtual clinical trials. The virtual patients are generated from retrospective MV patient data using a clinically validated respiratory mechanics model whose respiratory parameters (respiratory elastance and resistance) capture patient-specific pulmonary conditions and responses to MV care over time. Patient-level validation compares the predicted responses from the virtual patient to their retrospective results for clinically implemented MV settings and changes to care. Patient-level validated virtual patients create a platform to conduct virtual trials, where the safety of closed-loop model-based protocols can be evaluated.

RESULTS

This research creates and presents a virtual patient platform of 100 virtual patients generated from retrospective data. Patient-level validation reported median errors of 3.26% for volume-control and 6.80% for pressure-control ventilation mode. A virtual trial on a model-based protocol demonstrates the potential efficacy of using virtual patients for prospective evaluation and testing of the protocol.

CONCLUSION

The virtual patient framework shows the potential to safely and rapidly design, develop, and optimise new model-based MV decision support systems and protocols using clinically validated models and computer simulation, which could ultimately improve patient care and outcomes in MV.

Quantification of respiratory effort magnitude in spontaneous breathing patients using Convolutional Autoencoders

BACKGROUND

Spontaneous breathing (SB) effort during mechanical ventilation (MV) is an important metric of respiratory drive. However, SB effort varies due to a variety of factors, including evolving pathology and sedation levels. Therefore, assessment of SB efforts needs to be continuous and non-invasive. This is important to prevent both over- and under-assistance with MV. In this study, a machine learning model, Convolutional Autoencoder (CAE) is developed to quantify the magnitude of SB effort using only bedside MV airway pressure and flow waveform.

METHOD

The CAE model was trained using 12,170,655 simulated SB flow and normal flow data (NB). The paired SB and NB flow data were simulated using a Gaussian Effort Model (GEM) with 5 basis functions. When the CAE model is given a SB flow input, it is capable of predicting a corresponding NB flow for the SB flow input. The magnitude of SB effort (SBE_Mag) is then quantified as the difference between the SB and NB flows. The CAE model was used to evaluate the SBE_Mag of 9 pressure control/ support datasets. Results were validated using a mean squared error (MSE) fitting between clinical and training SB flows.

RESULTS

The CAE model was able to produce NB flows from the clinical SB flows with the median SBE_Mag of the 9 datasets being 25.39% [IQR: 21.87-25.57%]. The absolute error in SBE_Mag using MSE validation yields a median of 4.77% [IQR: 3.77-8.56%] amongst the cohort. This shows the ability of the GEM to capture the intrinsic details present in SB flow waveforms. Analysis also shows both intra-patient and inter-patient variability in SBE_Mag.

CONCLUSION

A Convolutional Autoencoder model was developed with simulated SB and NB flow data and is capable of quantifying the magnitude of patient spontaneous breathing effort. This provides potential application for real-time monitoring of patient respiratory drive for better management of patient-ventilator interaction.

Virtual patients for mechanical ventilation in the intensive care unit

BACKGROUND

Mechanical ventilation (MV) is a core intensive care unit (ICU) therapy. Significant inter- and intra- patient variability in lung mechanics and condition makes managing MV difficult. Accurate prediction of patient-specific response to changes in MV settings would enable optimised, personalised, and more productive care, improving outcomes and reducing cost. This study develops a generalised digital clone model, or in-silico virtual patient, to accurately predict lung mechanics in response to changes in MV.

METHODS

An identifiable, nonlinear hysteresis loop model (HLM) captures patient-specific lung dynamics identified from measured ventilator data. Identification and creation of the virtual patient model is fully automated using the hysteresis loop analysis (HLA) method to identify lung elastances from clinical data. Performance is evaluated using clinical data from 18 volume-control (VC) and 14 pressure-control (PC) ventilated patients who underwent step-wise recruitment maneuvers.

RESULTS

Patient-specific virtual patient models accurately predict lung response for changes in PEEP up to 12 cmH₂O for both volume and pressure control cohorts. R² values for predicting peak inspiration pressure (PIP) and additional retained lung volume, V_frc in VC, are R²=0.86 and R²=0.90 for 106 predictions over 18 patients. For 14 PC patients and 84 predictions, predicting peak inspiratory volume (PIV) and V_frc yield R²=0.86 and R²=0.83. Absolute PIP, PIV and V_frc errors are relatively small.

CONCLUSIONS

Overall results validate the accuracy and versatility of the virtual patient model for capturing and predicting nonlinear changes in patient-specific lung mechanics. Accurate response prediction enables mechanically and physiologically relevant virtual patients to guide personalised and optimised MV therapy.

Patient asynchrony modelling during controlled mechanical ventilation therapy

BACKGROUND AND OBJECTIVE

Mechanical ventilation therapy of respiratory failure patients can be guided by monitoring patient-specific respiratory mechanics. However, the patient's spontaneous breathing effort during controlled ventilation changes airway pressure waveform and thus affects the model-based identification of patient-specific respiratory mechanics parameters. This study develops a model to estimate respiratory mechanics in the presence of patient effort.

METHODS

Gaussian effort model (GEM) is a derivative of the single-compartment model with basis function. GEM model uses a linear combination of basis functions to model the nonlinear pressure waveform of spontaneous breathing patients. The GEM model estimates respiratory mechanics such as Elastance and Resistance along with the magnitudes of basis functions, which accounts for patient inspiratory effort.

RESULTS AND DISCUSSION

The GEM model was tested using both simulated data and a retrospective observational clinical trial patient data. GEM model fitting to the original airway pressure waveform is better than any existing models when reverse triggering asynchrony is present. The fitting error of GEM model was less than 10% for both simulated data and clinical trial patient data.

CONCLUSION

GEM can capture the respiratory mechanics in the presence of patient effect in volume control ventilation mode and also can be used to assess patient-ventilator interaction. This model determines basis functions magnitudes, which can be used to simulate any waveform of patient effort pressure for future studies. The estimation of parameter identification GEM model can further be improved by constraining the parameters within a physiologically plausible range during least-square nonlinear regression.

Optimising mechanical ventilation through model-based methods and automation

Mechanical ventilation (MV) is a core life-support therapy for patients suffering from respiratory failure or acute respiratory distress syndrome (ARDS). Respiratory failure is a secondary outcome of a range of injuries and diseases, and results in almost half of all intensive care unit (ICU) patients receiving some form of MV. Funding the increasing demand for ICU is a major issue and MV, in particular, can double the cost per day due to significant patient variability, over-sedation, and the large amount of clinician time required for patient management. Reducing cost in this area requires both a decrease in the average duration of MV by improving care, and a reduction in clinical workload. Both could be achieved by safely automating all or part of MV care via model-based dynamic systems modelling and control methods are ideally suited to address these problems. This paper presents common lung models, and provides a vision for a more automated future and explores predictive capacity of some current models. This vision includes the use of model-based methods to gain real-time insight to patient condition, improve safety through the forward prediction of outcomes to changes in MV, and develop virtual patients for in-silico design and testing of clinical protocols. Finally, the use of dynamic systems models and system identification to guide therapy for improved personalised control of oxygenation and MV therapy in the ICU will be considered. Such methods are a major part of the future of medicine, which includes greater personalisation and predictive capacity to both optimise care and reduce costs. This review thus presents the state of the art in how dynamic systems and control methods can be applied to transform this core area of ICU medicine.

Biomedical engineer's guide to the clinical aspects of intensive care mechanical ventilation

Background

Mechanical ventilation is an essential therapy to support critically ill respiratory failure patients. Current standards of care consist of generalised approaches, such as the use of positive end expiratory pressure to inspired oxygen fraction (PEEP–FiO₂) tables, which fail to account for the inter- and intra-patient variability between and within patients. The benefits of higher or lower tidal volume, PEEP, and other settings are highly debated and no consensus has been reached. Moreover, clinicians implicitly account for patient-specific factors such as disease condition and progression as they manually titrate ventilator settings. Hence, care is highly variable and potentially often non-optimal. These conditions create a situation that could benefit greatly from an engineered approach. The overall goal is a review of ventilation that is accessible to both clinicians and engineers, to bridge the divide between the two fields and enable collaboration to improve patient care and outcomes. This review does not take the form of a typical systematic review. Instead, it defines the standard terminology and introduces key clinical and biomedical measurements before introducing the key clinical studies and their influence in clinical practice which in turn flows into the needs and requirements around how biomedical engineering research can play a role in improving care. Given the significant clinical research to date and its impact on this complex area of care, this review thus provides a tutorial introduction around the review of the state of the art relevant to a biomedical engineering perspective.

Discussion

This review presents the significant clinical aspects and variables of ventilation management, the potential risks associated with suboptimal ventilation management, and a review of the major recent attempts to improve ventilation in the context of these variables. The unique aspect of this review is a focus on these key elements relevant to engineering new approaches. In particular, the need for ventilation strategies which consider, and directly account for, the significant differences in patient condition, disease etiology, and progression within patients is demonstrated with the subsequent requirement for optimal ventilation strategies to titrate for patient- and time-specific conditions.

Conclusion

Engineered, protective lung strategies that can directly account for and manage inter- and intra-patient variability thus offer great potential to improve both individual care, as well as cohort clinical outcomes.

Assessing mechanical ventilation asynchrony through iterative airway pressure reconstruction

BACKGROUND AND OBJECTIVE

Respiratory mechanics estimation can be used to guide mechanical ventilation (MV) but is severely compromised when asynchronous breathing occurs. In addition, asynchrony during MV is often not monitored and little is known about the impact or magnitude of asynchronous breathing towards recovery. Thus, it is important to monitor and quantify asynchronous breathing over every breath in an automated fashion, enabling the ability to overcome the limitations of model-based respiratory mechanics estimation during asynchronous breathing ventilation.

METHODS

An iterative airway pressure reconstruction (IPR) method is used to reconstruct asynchronous airway pressure waveforms to better match passive breathing airway waveforms using a single compartment model. The reconstructed pressure enables estimation of respiratory mechanics of airway pressure waveform essentially free from asynchrony. Reconstruction enables real-time breath-to-breath monitoring and quantification of the magnitude of the asynchrony (M_Asyn).

RESULTS AND DISCUSSION

Over 100,000 breathing cycles from MV patients with known asynchronous breathing were analyzed. The IPR was able to reconstruct different types of asynchronous breathing. The resulting respiratory mechanics estimated using pressure reconstruction were more consistent with smaller interquartile range (IQR) compared to respiratory mechanics estimated using asynchronous pressure. Comparing reconstructed pressure with asynchronous pressure waveforms quantifies the magnitude of asynchronous breathing, which has a median value M_Asyn for the entire dataset of 3.8%.

CONCLUSION

The iterative pressure reconstruction method is capable of identifying asynchronous breaths and improving respiratory mechanics estimation consistency compared to conventional model-based methods. It provides an opportunity to automate real-time quantification of asynchronous breathing frequency and magnitude that was previously limited to invasively method only.

Prediction of high airway pressure using a non-linear autoregressive model of pulmonary mechanics

Background

For mechanically ventilated patients with acute respiratory distress syndrome (ARDS), suboptimal PEEP levels can cause ventilator induced lung injury (VILI). In particular, high PEEP and high peak inspiratory pressures (PIP) can cause over distension of alveoli that is associated with VILI. However, PEEP must also be sufficient to maintain recruitment in ARDS lungs. A lung model that accurately and precisely predicts the outcome of an increase in PEEP may allow dangerous high PIP to be avoided, and reduce the incidence of VILI.

Methods and results

Sixteen pressure-flow data sets were collected from nine mechanically ventilated ARDs patients that underwent one or more recruitment manoeuvres. A nonlinear autoregressive (NARX) model was identified on one or more adjacent PEEP steps, and extrapolated to predict PIP at 2, 4, and 6 cmH₂O PEEP horizons. The analysis considered whether the predicted and measured PIP exceeded a threshold of 40 cmH₂O. A direct comparison of the method was made using the first order model of pulmonary mechanics (FOM(I)). Additionally, a further, more clinically appropriate method for the FOM was tested, in which the FOM was trained on a single PEEP prior to prediction (FOM(II)). The NARX model exhibited very high sensitivity (> 0.96) in all cases, and a high specificity (> 0.88). While both FOM methods had a high specificity (> 0.96), the sensitivity was much lower, with a mean of 0.68 for FOM(I), and 0.82 for FOM(II).

Conclusions

Clinically, false negatives are more harmful than false positives, as a high PIP may result in distension and VILI. Thus, the NARX model may be more effective than the FOM in allowing clinicians to reduce the risk of applying a PEEP that results in dangerously high airway pressures.

Incorporating bolus and infusion pharmacokinetics into the ICING insulin model

The ICING model has been successfully used to guide clinical decisions on insulin administration in critical illness. However, insulin pharmacokinetics in the ICING model can be improved to better describe both intravenous (IV) bolus and infusion insulin administration. Patient data from 217 Dynamic Insulin Sensitivity and Secretion Tests (DISST) and 36 Intravenous Glucose Tolerance Tests (IVGTT) from independent dietary intervention studies was used to fit model parameters to a model structure that conforms to known behaviour. The DISST tests measured both endogenous and exogenous IV insulin bolus responses, while the IVGTT measured exogenous IV insulin infusion dynamics. Unidentifiable parameters were given physiologically justified values, with knowledge on relative insulin clearance rates used to constrain parameter values. The resulting whole-cohort description was able to simultaneously describe both IV bolus and infusion dynamics, and improves ICING model descriptive capability. Improved infusion dynamics will allow better description of subcutaneous insulin, the insulin administration route favoured in outpatient care of diabetes.

Extrapolation of a non-linear autoregressive model of pulmonary mechanics

For patients with acute respiratory distress syndrome (ARDS), mechanical ventilation (MV) is an essential therapy in the intensive care unit (ICU). Suboptimal PEEP levels in MV can cause ventilator induced lung injury, which is associated with increased mortality, extended ICU stay, and high cost. The ability to predict the outcome of respiratory mechanics in response to changes in PEEP would thus provide a critical advantage in personalising and improving care. Testing the potentially dangerous high pressures would not be required to assess their impact. A nonlinear autoregressive (NARX) model was used to predict airway pressure in 19 data sets from 10 mechanically ventilated ARDS patients. Patient-specific NARX models were identified from pressure and flow data over one, two, three, or four adjacent PEEP levels in a recruitment manoeuvre. Extrapolation of NARX model elastance functions allowed prediction of patient responses to PEEP changes to higher or lower pressures. NARX model predictions were more successful than those using a well validated first order model (FOM). The most clinically important results were for extrapolation up one PEEP step of 2cmH₂O from the highest PEEP in the training data. When the NARX model was trained on one PEEP level, the mean RMS residual for the extrapolation PEEP level was 0.52 (90% CI: 0.47-0.57) cmH₂O, compared to 1.50 (90% CI: 1.38-1.62) cmH₂O for the FOM. When trained on four PEEP levels, the NARX result was 0.50 (90% CI: 0.42-0.58) cmH₂O, and was 1.95 (90% CI: 1.71-2.19) cmH₂O for the FOM. The results suggest that a full recruitment manoeuvre may not be required for the NARX model to obtain a useful estimate of the pressure waveform at higher PEEP levels. The methodology could thus allow clinicians to make informed decisions about ventilator PEEP settings while reducing the risk associated with high PEEP, and subsequent high peak airway pressures.

A polynomial model of patient-specific breathing effort during controlled mechanical ventilation

Patient breathing efforts occurring during controlled ventilation causes perturbations in pressure data, which cause erroneous parameter estimation in conventional models of respiratory mechanics. A polynomial model of patient effort can be used to capture breath-specific effort and underlying lung condition. An iterative multiple linear regression is used to identify the model in clinical volume controlled data. The polynomial model has lower fitting error and more stable estimates of respiratory elastance and resistance in the presence of patient effort than the conventional single compartment model. However, the polynomial model can converge to poor parameter estimation when patient efforts occur very early in the breath, or for long duration. The model of patient effort can provide clinical benefits by providing accurate respiratory mechanics estimation and monitoring of breath-to-breath patient effort, which can be used by clinicians to guide treatment.

Time-varying respiratory system elastance: a physiological model for patients who are spontaneously breathing

BACKGROUND

Respiratory mechanics models can aid in optimising patient-specific mechanical ventilation (MV), but the applications are limited to fully sedated MV patients who have little or no spontaneously breathing efforts. This research presents a time-varying elastance (E(drs)) model that can be used in spontaneously breathing patients to determine their respiratory mechanics.

METHODS

A time-varying respiratory elastance model is developed with a negative elastic component (E(demand)), to describe the driving pressure generated during a patient initiated breathing cycle. Data from 22 patients who are partially mechanically ventilated using Pressure Support (PS) and Neurally Adjusted Ventilatory Assist (NAVA) are used to investigate the physiology relevance of the time-varying elastance model and its clinical potential. E(drs) of every breathing cycle for each patient at different ventilation modes are presented for comparison.

RESULTS

At the start of every breathing cycle initiated by patient, E(drs) is < 0. This negativity is attributed from the E(demand) due to a positive lung volume intake at through negative pressure in the lung compartment. The mapping of E(drs) trajectories was able to give unique information to patients' breathing variability under different ventilation modes. The area under the curve of E(drs) (AUCE(drs)) for most patients is > 25 cmH2Os/l and thus can be used as an acute respiratory distress syndrome (ARDS) severity indicator.

CONCLUSION

The E(drs) model captures unique dynamic respiratory mechanics for spontaneously breathing patients with respiratory failure. The model is fully general and is applicable to both fully controlled and partially assisted MV modes.